Re: Hashing urls with Adler32

[Tomte]

It produces an output of fixed length for an input of arbitrary length. So there is an infinit set of possible inputs mapped to a finit set of checksums - so the algorithm can't produce a unique checksum for every url you feed it. I suggest you read the article on wikipedia and then have a look at SHA1 as possibly the better solution - that nonetheless will not be bijective either (It will produce collisions!) - it depends on your problem at hand if this is a hindrance.

[isync]

Currently I am using MD5 as digest, but with lots of urls the data structure is growing big.

So I thought about reducing the bits per url and using adler32 instead.

BTW: I am implementing a url-seen structure here and need the hash to check against, while minimizing false positives/negatives.